This PhD position is granted by the VISITOR European project. Candidates should check their eligibility and apply according to the guidelines in: http://www.inrialpes.fr/movi/pub/Visitor/

Real-time realistic simulation of rivers

Contact: Fabrice.Neyret@imag.fr
URL: http://www-evasion.imag.fr/Membres/Fabrice.Neyret/

keywords: image synthesis, fluids, CFD, hydraulics, phenomenological models

Our long-term goal is the real-time animation of very detailed rivers. Special effects for movies and video games use more and more fluids (water, ocean waves, explosions...). Still, the high-resolution CFD simulation they use is extremely costly, cannot afford very small details like capillary ripples or complex phenomena like foam, and lets little control for the artist. Conversely, to human eyes the aspect of the water surface of rivers is characterized by macroscopic features, which are exactly the handles an artist would have to design such a scene. It occurs that a pre-computer days physics as well as dedicated models (e.g. hydraulics) relying on such macroscopic entities do exist ! We explored some of it to produce examples of real-time high-resolution (vectorial, indeed) river features (see biblio).

As illustrated on the bottom images this topic is promising. Still, almost everything has yet to be done: how to extrapolate 2D empirical models to 3D ? How to mix empirical models (instabilities, eddies, ripple patterns) together, and to low resolution numerical fluids simulation ? How to handle efficiently the animation and rendering of very large scenes ? How to deal with highly complex but poorly known phenomenons such as foam or cascades ? How to introduce user interaction ? As one can see, there are avenues for interesting and ambitious multi-domains researches so to get impressive results !

Prerequisite: C/C++, OpenGL, Computer Graphics, notions on fluid physics (hydraulics, CFD).

References:
[NP01] "Phenomenological Simulation of Brooks”, F.Neyret, N.Praizelin, EG Computer Animation and Simulation, Sep 2001.
[DN01] "Simulation d'un ruisseau par approches phenomenologiques pour la synthese d'images", C.Dodard, RR, juin 2001.